Open AccessLung Cancer Diagnosis using Transfer Learning
Author(s) -
Aashka Mohite
Publication year - 2021
Publication title -
international journal of scientific research and management
Language(s) - English
Resource type - Journals
ISSN - 2321-3418
DOI - 10.18535/ijsrm/v9i11.ec02
Subject(s) - transfer of learning , lung cancer , recall , lung , convolutional neural network , artificial intelligence , medicine , cancer , computer science , radiology , pattern recognition (psychology) , pathology , psychology , cognitive psychology
Lung cancer is unquestionably a lung-influencing chronic condition that significantly hampers the respiratory system. It is the second most dangerous disease which causes increase in death rate. To resolve this issue, we had planned to create a very Convolutional Neural Network using Transfer learning to specifically classify the lungs CT scans as normal, malignant, or benign in a subtle way. A dataset of 1100 lung CT scans is used for this purpose. For the most part, five Transfer Learning architectures are compared extensively in this classification such as MobileNet, VGG16, VGG19, DenseNet-201 and ResNet-101. Out of which, DenseNet-201 performed the best. The proposed strategy achieved a mean accuracy of 53 percent in the trials and 43% of mean F1-score, mean precision and mean recall.